SPI bridge for AWA protocol to control a LED strip from HyperHDR (version v17 and above).
Diagnostic data available at the serial port output (@115200 speed).
Rpi acts as a master, ESP8266/ESP32 is in slave mode.
LED strip / Device | ESP8266 | ESP32 |
---|---|---|
SK6812 cold white | yes | yes |
SK6812 neutral white | yes | yes |
WS281x | yes | yes |
- SPI is much faster. HyperSPI works best at speed over 20Mb.
- SPI doesn't have any data integration check. But AWA protocol does have one.
- you don't need to have 2Mb capable serial port on your ESP board.
- SPI transmission is much lighter than serial communication
- I needed it and I was able to implemented it 😉
- There is a hardware limitation for the Rpi current design...even if you connect your grabber to the USB3.0 in the USB2.0 mode the adalight running driver causes quite a big USB transfer drop. So we can replace Adalight with a pure SPI data transfer as an alternative.
See what's happening for USB2.0 bus... my problematic Ezcap 320 @ 50 fps fell back to USB2.0 mode and did not like the CH340G serial port driver at all (real USB3.0 should not be affected, but not tested it):
That's how the grabbers works when other device is disconnected from the USB port.
In HyperHDR Image Processing→Smoothing→Update frequency
you should do not exceed the maximum capacity of the device. Read more here: testing performance
Select esp8266 protocol for ESP proprietary SPI protocol, esp32 for ESP32 boards or 'standard' for other devices.
Make sure you set "Refresh time" to zero, "Baudrate" should be set to high but realistic value like 25 000 000
.
Enabling "White channel calibration" is optional, if you want to fine tune the white channel balance of your sk6812 RGBW LED strip.
If you are using an ESP board compatible with the Wemos board (ESP8266 Wemos D1/pro, ESP32 MH-ET Live, ESP32-S2 lolin mini), the SPI connection uses the same pinout location on the ESP board! The pin positions of the LED output may vary. Cables (including ground) should not exceed 15-20cm or it may be necessary to lower the SPI speed. See how easy it is to connect ESP to Rpi:
ESP8266 | PINOUT |
---|---|
Clock (SCK) | GPIO 14 |
Data (MOSI) | GPIO 13 |
GROUND | mandatory |
LED output | GPIO 2 |
ESP32 | PINOUT |
---|---|
Clock (SCK) | GPIO 18 |
Data (MOSI) | GPIO 23 |
SPI Chip Select(e.g. CE0) | GPIO 5 |
GROUND | mandatory |
LED output | GPIO 2 |
ESP32-S2 lolin mini | PINOUT |
---|---|
Clock (SCK) | GPIO 7 |
Data (MOSI) | GPIO 11 |
SPI Chip Select(e.g. CE0) | GPIO 12 |
GROUND | mandatory |
LED output | GPIO 2 |
There are two versions of the firmware for ESP32 and ESP32-S2. The 'factory' and the 'base' one. Factory firmware should be flashed to offset 0x0, base firmware to offset 0x10000.
ESP32-S2 Lolin mini:
Requires using esptool.py
to flash the firmware e.g.
esptool.py write_flash 0x10000 hyperspi_esp32_s2_mini_SK6812_RGBW_COLD.bin
oresptool.py write_flash 0x0 hyperspi_esp32_s2_mini_SK6812_RGBW_COLD.factory.bin
Troubleshooting: ESP32-S2 Lolin mini recovery procedure.
- Put the board into dfu mode using board buttons: press board
Rst
+0
buttons, then releaseRst
, next release0
Do not reset or disconnect the board until the end of the recovery procedure. - Execute
esptool.py erase_flash
- Get circuitpython Execute
esptool.py write_flash 0x0 adafruit-circuitpython-lolin_s2_mini-pl-8.0.0.bin
- Execute
esptool.py write_flash 0x10000 hyperspi_esp32_s2_mini_SK6812_RGBW_COLD.bin
- Reset the board manually
Generic Esp8266/ESP32:
Recommend to use esphome-flasher
For RGBW LED strip like RGBW SK6812 NEUTRAL white choose: hyperspi_..._SK6812_RGBW_NEUTRAL.bin
For RGBW LED strip like RGBW SK6812 COLD white choose: hyperspi_..._SK6812_RGBW_COLD.bin
For RGB LED strip like WS8212b or RGB SK6812 variant choose: hyperspi_..._WS281x_RGB.bin
If you want to disable your first LED because it's used as a sacrificial level shifter, please use HyperHDR v19
For the RGBW firmware the white channel is automatically calculated and R,G,B channels are corrected.
LED strip / Device | ESP32 MH-ET LIVE mini HyperSPI v9 |
ESP32-S2 Lolin mini HyperSPI v9 |
---|---|---|
300LEDs sk6812 Refresh rate/continues output=100Hz SECOND_SEGMENT_START_INDEX=150 |
100 | 100 |
600LEDs sk6812 Refresh rate/continues output=83Hz SECOND_SEGMENT_START_INDEX=300 |
83 | 83 |
900LEDs sk6812 Refresh rate/continues output=55Hz SECOND_SEGMENT_START_INDEX=450 |
54-55 | 55 |
LED strip / Device | ESP32 MH ET Live HyperSPI v9 |
---|---|
300LEDs RGBW Refresh rate/continues output=83Hz |
83 |
600LEDs RGBW Refresh rate/continues output=43Hz |
42-43 |
900LEDs RGBW Refresh rate/continues output=28Hz |
28 |
LED strip / Device | ESP32-S2 Lolin mini HyperSPI v9 |
---|---|
300LEDs RGBW Refresh rate/continues output=83Hz |
83 |
600LEDs RGBW Refresh rate/continues output=43Hz |
42 |
900LEDs RGBW Refresh rate/continues output=28Hz |
28 |
LED strip / Device | ESP8266 Wemos D1 Pro HyperSPI v9 |
---|---|
300LEDs RGBW Refresh rate/continues output=70Hz |
70 |
600LEDs RGBW Refresh rate/continues output=33Hz |
33 |
900LEDs RGBW Refresh rate/continues output=22Hz |
22 |
Esp8266 Wemos D1 mini (CH340) and Wemos D1 mini pro (CP2104)
ESP32 MH-ET Live and ESP32-S2 Lolin mini (CDC)
Currently we use PlatformIO to compile the project. Install Visual Studio Code and add PlatformIO plugin. This environment will take care of everything and compile the firmware for you. Low-level LED strip support is provided by my highly optimizated (pre-fill I2S DMA modes, turbo I2S parallel mode for up to 2 segments etc) version of Neopixelbus library: link.
But there is also an alternative and an easier way. Just fork the project and enable its Github Action. Use the online editor to make changes to the platformio.ini
file, for example change default pin-outs or enable multi-segments support, and save it. Github Action will compile new firmware automatically in the Artifacts archive. It has never been so easy!
Tutorial: https://github.com/awawa-dev/HyperSPI/wiki
Using parallel multi-segment allows you to double your Neopixel (e.g. sk6812 RGBW) LED strip refresh rate by dividing it into two smaller equal parts. Both smaller segments are perfectly in sync so you don't need to worry about it. Proposed example of building a multisegment:
- Divide a long or dense strip of LEDs into 2 smaller equal parts. So
SECOND_SEGMENT_START_INDEX
in the HyperSPI firmware is the total number of LEDs divided by 2. - Build your first segment traditional way e.g. clockwise, so it starts somewhere in middle of the bottom of frame/TV and ends in the middle of the top of frame/TV
- Start the second segment in the opposite direction to the first one e.g. counterclockwise (
SECOND_SEGMENT_REVERSED
option in the HyperSPI firmware configuration must be enabled). So it starts somewhere in the middle of the bottom of the frame/TV and ends in the middle of the top of the TV/frame. Both segments could be connected if possible at the top but only 5v and ground ( NOT the data line). - The data line starts for both segments somewhere in the middle of the bottom of the TV/frame (where each of the LED strips starts)
- Configuration in HyperHDR does not change! It's should be configured as one, single continues segment. All is done in HyperSPI firmware transparently and does not affect LED strip configuration in HyperHDR.
You also must configure data pin in the platformio.ini
. Review the comments at the top of the file:
SECOND_SEGMENT_DATA_PIN
- These is data pin for your second strip
You add these to your board's config. Be sure to put -D
in front of each setting.
Examples of final build_flags for 288 LEDs divided into 2 equal segments in the platformio.ini
:
[env:SK6812_RGBW_COLD]
build_flags = -DNEOPIXEL_RGBW -DCOLD_WHITE -DDATA_PIN=2 ${env.build_flags} -DSECOND_SEGMENT_START_INDEX=144 -DSECOND_SEGMENT_DATA_PIN=4 -DSECOND_SEGMENT_REVERSED
...
[env:WS281x_RGB]
build_flags = -DNEOPIXEL_RGB -DDATA_PIN=2 ${env.build_flags} -DSECOND_SEGMENT_START_INDEX=144 -DSECOND_SEGMENT_DATA_PIN=4 -DSECOND_SEGMENT_REVERSED
...
Implementation example:
- The diagram of the board for WS2812b/SK6812 including ESP32 and the SN74AHCT125N 74AHCT125 level shifter.
The output is only available when HyperHDR is not using the device at the moment, so it should be disabled in the app for a while. Stores the last result when HyperHDR was running in the current session. You can read it from the serial port at a speed of 115200.
For testing maximum performance in HyperHDR enable Image Processing→Smoothing→Continuous output
, high Update frequency
in the same tab and set any color in the Remote control
tab as an active effect. After testing you need to disable Continuous output
and set Update frequency
according to your results.